Network dismantling aims to find the minimal set of nodes that, if removed, will break the network into small components with their largest one limited to a certain threshold. This problem underlies many practical applications in various areas, such as bioinformatics, transportation and the Internet. There are two major kinds of network dismantling methods, i.e. centrality measure based methods and network decycling based methods. The former ignores the influence of the loop structure in network topology, while the latter massively deletes irrelevant nodes in the loop removal step, both resulting in poor performance. To solve these problems, this paper proposes a neighborhood link sensitive dismantling method for social networks. The proposed method contains two key steps, namely node deleting step and node re-inserting step. In node deleting step, a neighborhood link sensitive centrality measure is defined to identify the nodes that are really crucial to destroy the connectivity of network. In the following node re-inserting step, an appropriate greedy strategy is selected to refine the node set of network dismantling as much as possible to the theoretical optimal solution. Experimental results on real-world networks and synthetic networks demonstrate that the proposed method can break down networks by deleting only a smaller set of nodes, outperforming the existing state-of-the-art methods. Furthermore, our proposed method shows stable performance and strong adaptability on networks with different scales and structural characteristics.

网络拆卸的目的是找到最小的节点集，如果删除这些节点，它们会将网络分解成小的组件，而最大的组件则限制在某个阈值内。这个问题是生物信息学，运输和互联网等各个领域的许多实际应用的基础。网络拆卸方法主要有两种，即基于中心度度量的方法和基于网络回收的方法。前者忽略了环路结构在网络拓扑结构中的影响，而后者则在环路去除步骤中大量删除了不相关的节点，均导致性能下降。为了解决这些问题，本文提出了一种用于社交网络的邻域链接敏感拆除方法。所提出的方法包括两个关键步骤，即节点删除步骤和节点重新插入步骤。在节点删除步骤中，定义了邻域链路敏感度集中度度量，以标识对于破坏网络连接性至关重要的节点。在接下来的节点重新插入步骤中，选择适当的贪婪策略以将网络拆卸的节点集尽可能地细化为理论上的最佳解决方案。在实际网络和合成网络上的实验结果表明，所提出的方法可以通过仅删除较小的节点集来破坏网络，从而胜过现有的最新技术。此外，我们提出的方法在具有不同规模和结构特征的网络上表现出稳定的性能和较强的适应性。